© 2003 Kluwer Academic Publishers. Printed in the Netherlands.
NMR of bicelles: orientation and mosaic spread of the liquid-crystal
director under sample rotation
Giorgia Zandomeneghi, Marco Tomaselli, Philip T. F. Williamson & Beat H. Meier
∗Physical Chemistry, ETH Zurich, ETH-Hönggerberg, CH-8093 Zurich, Switzerland
Received 21 August 2002; Accepted 31 October 2002
Key words: bicelle, liquid-crystal, NMR, order parameter, variable-angle spinning
Abstract
Model-membrane systems composed of liquid-crystalline bicellar phases can be uniaxially oriented with respect to a magnetic field, thereby facilitating structural and dynamics studies of membrane-associated proteins. Here we quantitatively characterize a method that allows the manipulation of the direction of this uniaxial orientation. Bicelles formed from DMPC/DHPC are examined by 31P NMR under variable-angle sample-spinning (VAS) conditions, confirming that the orientation of the liquid-crystalline director can be influenced by sample spinning. The director is perpendicular to the rotation axis when (the angle between the sample-spinning axis and the magnetic field direction) is smaller than the magic angle, and is parallel to the rotation axis when is larger than the magic angle. The new31P NMR VAS data presented are considerably more sensitive to the orientation of the bicelle than earlier2H studies and the analysis of the sideband pattern allows the determination of the orientation of the liquid-crystal director and its variation over the sample, i.e., the mosaic spread. Under VAS, the mosaic spread is small if deviates significantly from the magic angle but becomes very large at the magic angle.
Introduction
Bicelles self-organize from a mixture of long- and short-chain phospholipids in water (Sanders and Schwonek, 1992; Sanders et al., 1994; Sanders and Landis, 1995; Vold and Prosser, 1996). Liquid-crystalline phases consisting of bicelles are formed if the total lipid concentration, the ratio of the two phos-pholipids, the temperature and the pH of the solution are in the appropriate range. In this phase the phos-pholipids are assumed to form bilayered disks with a planar bilayer of long-chain DMPC phospholipids surrounded by a rim of shorter-chain DHPC (Vold and Prosser, 1996). Recently, this model has been ques-tioned and a model where DMPC constitutes an highly dynamic perforated bilayer and DHPC forms the rims surrounding the holes has been proposed (Gaemers and Bax, 2001; Nieh et al., 2001, 2002). For the interpretation of our data, both models are equiva-∗To whom correspondence should be addressed. E-mail: [email protected]
lent but in our graphical representations we use the standard model. In a magnetic field, B0, the
liquid-crystalline phase becomes oriented with the director perpendicular to B0. This particular director
orienta-tion is a consequence of the negative anisotropy of the diamagnetic susceptibility tensor, χ.
Bicelles are attractive as model membranes for the study of phospholipid-associated proteins (Sanders and Landis, 1995; Glover et al., 2001; Howard and Opella, 1996; Losonczi and Prestegard, 1998; Struppe et al. 1998). Well-oriented liquid-crystalline phases can represent a viable alternative to mechanically ori-ented bilayers on glass plates (Marassi et al., 1999; Moll and Cross, 1990). In experiments with bicellar systems, in particular for the study of membrane-bound proteins, it is interesting to manipulate the orientation of the liquid-crystalline director e.g. by sample-spinning techniques. For spinning frequencies larger than a critical frequency, which depends on the strength of B0, χ and the Leslie viscosity
co-efficient, the director orients such that the magnetic energy averaged over a rotor cycle,
Figure 1. Energetically most favorable orientation of the bicelle
di-rector with respect to the rotor axis at different angles between the rotor axis and B0.
Emag(,β) = −
χ
3µ0
B20· P2(cos )· P2(cosβ),
(1) is minimized (Courtieu et al., 1982, 1994). Here, is the angle between the magnetic field and the spin-ning axis νr and β is the angle between the bicelle director andνr. The vacuum permeability is denoted byµ0and the second-order Legendre polynomial by
P2(cosθ) = (3 cos2θ − 1)/2.
Two regimes of orientational behavior can be dis-tinguished, as illustrated in Figure 1. In the range 0◦≤
≤ m(where mis the magic angle, m≈ 54.7◦)
the energy is minimized when the bicelle director is perpendicular to the rotation axis (β = 90◦) (Tian
et al., 1999); for the range m < ≤ 90◦, the
min-imum corresponds to the parallel orientation (β = 0◦)
(Zandomeneghi et al., 2001). At the magic angle, the magnetic energy is independent of the director orien-tation and the bicelles are predicted from Equation 1 to be randomly oriented.
Our earlier work focused on the residual quadrupo-lar splitting of the deuterium resonance of the solvent water (Zandomeneghi et al., 2001). The dependence of the orientation of the director on the angle as ex-pressed by Equation 1, was experimentally confirmed for spinning frequencies between 400 and 800 Hz. For angles m< ≤ 90◦, especially at higher spinning
frequencies (above 900 Hz), an additional phase was detected and assigned to an orientation withβ = 90◦. Furthermore, it was shown that the director can be aligned with B0 by switched-angle spinning (SAS)
techniques. This orientation is spectroscopically at-tractive because it results in a single resonance line per site and therefore improves the spectral resolution (Howard and Opella, 1996; Prosser et al., 1998).
A number of different SAS experiments can be en-visioned to correlate spectra at different orientations or to distinguish between scalar and dipolar couplings. In a companion paper (Zandomeneghi et al., 2003) we describe the realization of SAS experiments applied to bicelles containing phospholipid-associated peptides.
Here we report a31P NMR study of the effect of sample spinning on the magnetic orientation of bicel-lar systems. We show that 31P spectroscopy can be used to describe not only the most likely orientation of the director but also the mosaic spread. This al-lows us also to characterize the additional phase that is sometimes observed in particular at higher spinning frequency or for angles close to the magic angle.
The31P spectra of oriented bicelles
The spin interaction observed in the1H-decoupled31P NMR spectrum is the anisotropic chemical-shielding interaction. The frequencyω in the NMR spectrum is determined by σ, the zz component of the cartesian CSA tensor in the laboratory frame,σL, according to
ω = −σω0= −(σL)zzω0, (2)
whereω0is the Larmor frequency.
Chemical-shielding-anisotropy (CSA) tensors will be characterized by the isotropic value ¯σ, the anisotropyδ, and the asymmetry-parameter η which are defined in terms of the elements of the cartesian matrix representation as ¯σ = (σxx + σyy + σzz)/3,
δ = σzz− ¯σ and η = (σxx− σyy)/δ.∗ In the principal
axis system, the CSA tensor has the representation:
σC = σ0xx σ0yy 00 0 0 σzz = ¯σ 1 0 00 1 0 0 0 1 + δC 2 −10 −1 00 0 0 0 2 +1 2δ CηC 10 −1 00 0 0 0 0 . (3)
∗The three principal values of the space tensor are ordered according to the convention:|σzz− ¯σ| ≥σyy− ¯σ ≥ |σxx− ¯σ|
Figure 2. Summary of coordinate frames employed to calculate the 31P spectra.
To calculate the spectrum, the CSA tensor in the laboratory frame must be known. This can be obtained through the series of reference frame transformations outlined in Figure 2. Each transformation is described by a set of three Euler angles, denoted by =
(α, β, γ).
The first transformation is from the principal axis system (CSA frame) located at the phosphorous nu-cleus to the bicelle frame, with its z-axis normal to the DMPC plane in the bicelle. The Euler angles for this transformation (t)CBare stochastically time-dependent due to internal rotations, overall molecular rotation and molecular diffusion (Kohler and Klein, 1977; Seelig, 1978; Dufourc et al., 1992). Signifi-cantly different behavior between DMPC and DHPC is expected for the molecular diffusion. In both mod-els for bicelles DMPC diffuses in a plane, leavingσ invariant. DHPC, in contrast, diffuses over a curved surface which could be modelled as half of a torus (Vold and Prosser, 1996; Picard et al., 1999).
The second transformation leads to the director frame, with its z-axis along the director of the par-ticular liquid-crystalline domain. The corresponding set of Euler angles (t)BDis again stochastically time-dependent due to dynamic deviations of an individual
bicelle normal from the direction of the director of the liquid crystalline phase. In the time average, these angles vanish: (t)BD = (0, 0, 0).
If the correlation time of the motions creating the time dependence of (t)CBand (t)BDis short com-pared to the NMR timescale, defined by (δCω
0)−1and
in the order of 100µs, the NMR spectrum from each domain is described by an averaged tensorσD, char-acterized by ¯σ and δD. Due to the axial symmetry, we expectηD = 0 and the unique axis of σD to lie along the liquid-crystalline director. This behavior is indeed found in our experiments (vide infra) for both DHPC and DMPC. Due to the geometrical differences in the motion of DMPC and DHPC, the anisotropy, δD, is quite different for the two components (Vold and
Prosser, 1996).
The anisotropy in the director frame is related to the anisotropy in the principal axis system by
δD = S Bic· δC (3 cos2βCB− 1) 2 +ηC 2 sin 2βCBcos 2αCB . (4)
The brackets denote time-averages and SBic =
3 cos2βBD− 1 /2 is the order parameter (Saupe,
1964) associated with the bicelle motion around the director.
In the following step, the CSA tensor must be transformed from the director frame (of each domain) to the rotor frame, a coordinate system fixed to the rotor (the sample container) with its z-axis parallel to the rotation axis.
This transformation is described by the Euler ro-tation DR. Due to symmetry, only two Euler angles are needed, i.e., DR = (0, β, γ). In general, β and γ do not assume a single value but are characterized by probabilities p(<β) and p(γ) because the sample can consist of several domains, each of them containing a large number of bicelles. Each domain can have a different director orientation with respect to the rotor frame. In particular, γ will be randomly distributed, because the energy in Equation 1 depends only onβ. In general, the angleβ will also be distributed and we will call this distribution the mosaic spread. We assume, that the director of each domain remains constant on the NMR timescale and that bicelles do not diffuse from one domain to the other. Therefore, this trans-formation does not involve time averaging and leaves ¯σ, δ and η invariant.
The final step, from the rotor frame to the laboratory frame, is described by the periodi-cally time-dependent transformation, RL(t) =
(−2πνrt,−, 0), where νris the rotor frequency and
the angle between the rotor axis and the magnetic field.
Transforming the bicelle-frame CSA tensor to the laboratory frame leads to the following expression for σ = (σL)
zz:
σ(νr, ,β, γ, t) = ¯σ + δD· P2(cosβ) · P2(cos )
+ (νr, ,β, γ, t). (5)
The time-dependent terms due to sample rotation are collected in the following relation:
(νr, ,β, γ, t) = C1(β, ) cos(2πνrt+ γ) + C2(β, ) cos(4πνrt+ 2γ), (6) with C1(β, ) = −34δDsin(2β) sin(2) (7) and C2(β, ) = 34δDsin(β)2sin()2. (8)
For a well-defined value of the angle β (corre-sponding to a vanishing mosaic spread of the liquid-crystal director) the spectrum consists of two families of sharp resonance lines for DHPC and DMPC, each consisting of a central line flanked by spinning side-bands separated by integer multiples ofνr. The
center-band resonance position depends on the angles and β as follows:
ωCB(β, ) = −(¯σ + δD· P2(cosβ) · P2(cos ))ω0.
(9) For a uniform distribution of the angleγ, p(γ) = 1/2π, all spinning sidebands are positive absorption lines with the intensity of the Nth sideband described by (Mehring, 1983) IN(β, )= 2π1 2π 0 exp i −Nφ −ω0C1(β, ) 2πνr sin(φ) −ω0C2(β, ) 4πνr sin(2φ) dφ 2 . (10)
We characterize the mosaic spread by a Gaussian distribution of the angle β centered at ¯β and with standard deviationµ, p(β) = 1 µ√2π· exp −1 2 β − ¯β µ 2 . (11)
Figure 3. Simulation of DMPC31P NMR sideband spectra for (a)
= 35◦and ¯β = 90◦andµ = 0◦, (b) = 66◦and ¯β = 0◦ andµ = 0◦, (c) = 35◦and ¯β = 90◦andµ = 7◦, and (d)
= 66◦and ¯β = 0◦andµ = 7◦. The highest peak is the center-band. In (a) only even order sidebands are observed, (b) shows no sidebands. The mosaic spread leads to the presence of even and odd order sidebands in (c) as well as in (d). For the anisotropy the value δD
DMPC= −22.0 ppm was used. Isotropic chemical shielding was ¯σ = 0 ppm. Note that ωCB(90◦, 35◦)≈ ωCB(0◦, 66◦).
For a non-vanishing mosaic spread, the center-and sidebcenter-ands will be heterogeneously broadened into a tensor pattern. The NMR spectrum is calculated according to S(,ω) =1 2· π 0 ∞ N=−∞ IN(β, ) · δD(ω− ωCB(β, ) − 2πNνr) p(β) · sin β dβ,(12) whereδDdenotes the Dirac delta function. Finally, a
Lorentzian line broadening can be added.
The resulting spinning-sideband patterns in the NMR spectrum are sensitive to ¯β and to the mosaic spread µ. We note that a perfectly oriented sample with ¯β = 90◦would give only even order sidebands while a perfectly oriented sample with ¯β = 0◦would give no sidebands at all, as shown in Figures 3a and 3b, respectively. In Figures 3c and 3d, the correspond-ing calculated spectra for a mosaic spread ofµ = 7◦ are given. They illustrate the strong sensitivity of the sideband intensities toµ.
Static spectra can be described with the same for-malism (Figure 2). The Euler angles DL = (0, β, γ) define the transformation from the director frame to
the laboratory frame. Again we use the Gaussian dis-tribution of Equation 11 with ¯β = 90◦and a random distribution forγ.
Materials and methods
Sample preparation
Fifty milligrams of lipids (1,2-Dihexanoyl-sn-glycero-3-phosphocholine (DHPC) and
1,2-Ditetradecanoyl-sn-glycero-3-phosphocholine (DMPC), [DMPC]/[DHPC]
= 3.5) were mixed with 150 µl of H2O. The following
procedure was repeated 3 times: vortexing (2 min), heating to 40◦C (20 min), vortexing (2 min), cooling to 0◦C (20 min). Lipids were purchased from Avanti Polar Lipids (Alabaster, AL). All experiments were performed with freshly prepared samples.
NMR experiments
31P NMR experiments were performed on a
home-built spectrometer (Hediger et al., 1997) at a mag-netic field of 7.00 Tesla (31P resonance frequency of
ω0/2π = −120.66 MHz), with a Chemagnetics MAS
probehead, using 4 mm rotors (internal diameter is 2.5 mm). The probehead was modified in order to en-large the range of accessible angles to 25◦–90◦. We used 90◦pulses of 3–4.5µs. The spectral width was 20 kHz and the 1H decoupling strength between 24 and 39 kHz. For each spectrum between 1000 and 4000 transients were acquired with a recycle time of 3 s. All measurements were done at 37◦C. The rota-tion angles have been measured by an inclinometer with an experimental uncertainty of 2.5◦. The magic angle was calibrated on the79Br signal of a spinning sample of polycrystalline KBr.
Simulations and calculations
Spectra have been processed with MATNMR.∗ The lineshape simulations were calculated in C++, us-ing the GAMMA spin-simulation environment (Smith et al., 1994). The powder averaging was performed according to Cheng (Cheng et al., 1973) with ei-ther 538 or 1154 powder orientations per spectrum. The least square fitting was performed with the pro-gram MINUIT.∗∗ The errors here reported are the ∗MATNMR is a toolbox for processing NMR/EPR data under MATLAB and can be downloaded freely at http:// www.nmr.ethz.ch/matnmr
∗∗MINUIT, Cern Program Library Entry D506.
Figure 4. (a) Experimental 31P NMR spectrum obtained under magic-angle spinning atνr = 400 Hz and (b) difference between the experimental and the fit described in the text. The spectrum reported has been measured 14 hours after spinning was started. The frequency scale is given in units of the chemical shielding, as defined in Equation 2.
ones obtained by MINUIT and represent one standard deviation.
Results and discussion
MAS spectra
The experimental 31P magic-angle spinning (MAS) spectrum of a bicellar phase of DMPC/DHPC is shown in Figure 4a. It consists of a single centerband (marked by a square) and the corresponding side-band family, indicating that the isotropic shieldings of DMPC (¯σDMPC) and DHPC (¯σDHPC) are identical
within our spectral resolution. In the following, we will reference all spectra to this isotropic shielding at the given temperature, setting ¯σDMPC = ¯σDHPC =
¯σ = 0 ppm. Referencing to 85% H3PO4leads, at T
= 310 K, to ¯σ ≈ 0.84 ppm.
At the magic angle, m ≈ 54.7◦, the magnetic
energy does not depend on (Equation 1) and an isotropic distribution of the director orientation is ex-pected. We will refer to such an isotropic distribution as a powder distribution. Note that powder distribution means that the director of different liquid-crystalline domains (each containing a large number of bicelles) is randomly oriented. The individual bicelles within one domain are still well-aligned with the director in this domain. The domains size is considerably larger
Figure 5. 31P NMR spectrum of a static sample.
than the free mean path of the solvent water mole-cules during the duration of the NMR experiment (milliseconds), otherwise the 2H splittings that have been observed (Zandomeneghi et al., 2001) would be absent. The statistic of the momentary misalignment of the bicelles with the director is described by the order parameter SBic.
The experimental sideband intensities in the 31P spectrum of Figure 4a, obtained after spinning a sam-ple at 400 Hz for 14 hours, are well described by a powder distribution of the director. Figure 4b shows the difference between the experimental data and a fit by a powder distribution for both phospholipids with the fitted anisotropiesδDDMPC= −21.7 ± 0.6 ppm and δD
DHPC= −7.9 ± 0.5 ppm. For shorter spinning times
the director showed some preferred ordering, oriented perpendicular to the rotation axis.
Static samples
In Figure 5 the 31P NMR spectrum of a static bi-cellar sample is reported. The spectrum consists of two sharp lines with resonance positions (relative to isotropic DHPC) of 11.25 ppm (DMPC) and 4.53 ppm (DHPC). The ratio of the peak integrals is, within experimental error, equal to the concentration ratio q = [DMPC]/[DHPC] = 3.5. It is well established (Sanders and Schwonek, 1992) that ¯β = 90◦for static samples. Therefore, the chemical shielding is
σ = ¯σ −δD
2 . (13)
The presence of one single sharp resonance line per compound corroborates that the chemical shielding tensors σD of both phospholipids are indeed axi-ally symmetric. We have determined δDDMPC, δDDHPC (according to Equation 13) and q (from the peak integrals) in the temperature range from 304 K to 314 K. At 310 K, we found δDDMPC = −23.0 ± 0.4 ppm, δDDHPC = −9.2 ± 0.4 ppm and q = 3.4 ± 0.3. The error margins given above are the standard deviation from a series of measurements on 20 different samples and are indicative of a certain variability on details of the sample preparation, ex-perimental temperature and sample history. Between 304 K to 314 K, the anisotropy of both components increases linearly with temperature, as described by linear coefficients dδD
DMPC/dT= −0.19 ppm/K and
dδDDHPC/dT= −0.46 ppm/K.
From a lineshape-analysis of the DHPC and DMPC signals, we can determine the mosaic spread to beµ = 4◦± 2◦.
Variable-angle spinning
31P NMR single pulse experiments were performed
on bicelle samples spun at different angles , us-ing spinnus-ing frequencies in the range 400 Hz ≤ νr
≤ 800 Hz. A selection of spectra taken at 800 Hz is given in Figure 6. From the sharp features of the spec-tra, tentatively assigned to a well-ordered phase, the two spinning-frequency independent centerbands of DHPC and DMPC, as well as the associated spinning-sideband families have been identified. A rigorous justification for this assignment will be presented be-low. The center-band resonance frequency is plotted, as function of in Figure 7. The solid lines in the figure correspond to a fit of these data using Equation 9 withβ = 90◦for < mandβ = 0◦for > m.
The fitted parameters are ¯σDMPC = 0.0 ± 0.2 ppm,
¯σDHPC = −0.1±0.1 ppm, δDDMPC= −22.7±0.8 ppm
and δDDHPC = −9.3 ± 0.4 ppm. The anisotropies are found to be identical to the ones determined in the static experiment within the error margins, and the isotropic shieldings are found to be zero. The good agreement indicates that (i) the anisotropy δD for DHPC and DMPC (and therefore the order para-meter of the liquid-crystalline phase) is not influenced by the spinning at frequencies used here. (ii) For an-gles smaller than the magic angle, the director of the well-ordered majority phase is perpendicular to the spinning axis, while for angles larger than the
Figure 6. 31P NMR spectra for different spinning angles , obtained at a spinning frequencyνr= 800 Hz. Column 1: Experimental spectrum; column 2: Best fit (bottom) and difference between the experimental spectrum and the fit (top); column 3: Contribution to the fit from DMPC (bottom) and DHPC (top) in the oriented phase; column 4: Contribution to the fit from DMPC (bottom) and DHPC (top) in the powder phase. Where indicated, intensities are multiplied by 3, 5 or 10. After the start of the sample’s spinning the evolution of the order was followed for about 12 h and the spectrum corresponding to the most ordered situation is reported here.
Figure 7. 31P NMR resonance positions for bicelle samples spun around an axis tilted by angle with respect to B0. Experimental points (♦), () and (∗) refer to DMPC peak, in experiments where νr = 800 Hz, νr = 400 Hz and νr = 550, 600 Hz, respectively. Experimental points (), () and () refer to DHPC peak, in exper-iments whereνr= 800 Hz, νr= 400 Hz and νr= 550, 600 Hz, re-spectively. Curve (a) describesσ = −0.1−P2(cos ¯β)·9.3·P(cos ), while Curve (b) representsσ = 0.0 − P2(cos ¯β) · 22.7 · P2(cos ). For 0◦≤ < m, P2(cos ¯β) = −0.5 and for m < ≤ 90◦, P2(cos ¯β) = 1.0. Angles are given as evaluated by an inclinome-ter and the error bars in the chemical shieldings are smaller than the width of the symbols.
magic angle, the two vectors are parallel, as shown schematically in Figure 1.
For a more detailed evaluation of the experimental spectra, we have fitted each spectrum using Equa-tion 12. The equaEqua-tion describes the spectrum for one compound, DHPC or DMPC, in a certain bicellar phase. We model our spectra as two different phases of bicelles, each one containing DMPC and DHPC and characterized by the stoichiometric ratio q, the anisotropy δD of DHPC and DMPC, the linewidth of the signals of DHPC and DMPC, and the mosaic spreadµ. For the first phase, ¯β = 90◦for < m
and ¯β = 0◦ for > m. For the second phase,
¯β = 90◦for all angles (Bayle et al., 1988), in
agree-ment with2H data for solvent water (Zandomeneghi et al., 2001). For νr < 850 Hz, the mosaic spread
for the minority phase was always found to be rather large (µ > 30◦) and it turned out, that spectra can
be equally well described by a simpler model where the more disordered phase is represented by a random distribution of the director with respect to B0, µ →
∞. In the following, this randomly oriented phase is called the ‘powder phase’. This model provides an excellent description of all experimental spectra with νr<850 Hz. Table 1 lists the parameters
correspond-ing to the best fit of the experimental data of Figure 6 (column 1) and of further spectra not shown. Figure 6 shows the fitted spectra and the difference between experiment and fit in column 2. The four different con-tributions to the calculated spectra, namely the DHPC and DMPC spectra in the oriented and powder phase respectively, are shown in columns 3 and 4. Good agreement between fit and experiments is found. Free fitting parameters are those listed in Table 1, as well as a Lorentzian line broadening (different for DMPC and DHPC) and an overall intensity scaling. The spin-ning angle has been set with an estimated precision of±2.5◦. However, it can also be used as a free fit parameter, despite the fact that δD and have sig-nificant correlation. The deviation between manually determined and fitted value was consistently below ±2.5◦.
It can be seen from column 3 of Figure 6 that the oriented phase leads, as expected, to a sharp sideband pattern for each component at all spinning angles . The anisotropies δDDHPC and δDDMPC are independent of (see Table 1). This finding fully corroborates the centerband-frequencies extracted directly from the spectra and plotted in Figure 7. The composition of the ordered phase, q, is always found to be around 3.5, as expected.
For angles away from the magic angle (e.g., < 45◦ or > 67◦) the mosaic spread of the bicel-lar phase is found to be around 7◦. This is slightly higher than in static samples and similar to mechan-ically oriented lipid bilayers on glass plates where mosaic spreads of 4◦ (Glaubitz and Watts, 1998) for pure DMPC and 8◦for a DMPC sample containing an integral membrane peptide have been reported (Moll and Cross, 1990).
Except at the magic angle, the powder phase is always the minority phase at the spinning speeds con-sidered here. At angles close to 0◦and 90◦, its weight becomes negligibly small. When the proportion of the powder phase is high enough to obtain precise fit values, the anisotropiesδD for both the phospho-lipids and the value of q are comparable with the ones found for the ordered phase and for the static sam-ples. This finding indicates that the disordered phase is also composed of bicelles, with similar DMPC/DHPC composition, shape and dynamics.
The experimental values for µ and the relative weight of the two phases are somewhat influenced by the details of the sample history. For < 45◦ or
> 67◦ the system assumes a constant degree of order after one hour, or even less, and this order starts
Table 1. δDvalues and ratio q= [DMPC]/[DHPC] for the oriented as well as for the powder phase, mosaic spread of the bicellar director’s orientation,µ and amount of the powder phase determined in VAS experiments where the sample is spun at angles with respect to B0and at spinning frequencyνr
(◦) νr Oriented phase Powder phase
(Hz) δDDMPC δDDHPC q µ (◦) Powder δDDMPC δDDHPC q (ppm) (ppm) phase (ppm) (ppm) (%) 26 (1) 800 −20.9(5) −8.3(2) 3.4(1) 7.3 (3) 17(12) −21.5(6) −7.8(4) 8 ( 5) 36.1(2) 400 −21.7(2) −8.0(1) 3.6(1) 6.8 (3) 7( 4) −20.4(4) −7.1(8) 10 ( 7) 35.8(4) 800 −21.6(5) −8.3(2) 3.6(2) 7.0 (4) 8( 3) −22.6(5) −8.9(4) 8 ( 6) 43.9(1) 400 −23.1(1) −8.8(1) 3.8(2) 5.9 (2) 20( 4) −21.5(2) −7.3(2) 3.1( 7) 43.8(2) 800 −21.8(4) −7.7(1) 3.1(1) 8.5 (6) 36( 7) −23.4(4) −6.5(5) 12 ( 7) 48.3(1) 430 −22.6(4) −9.2(2) 3.4(3) 8.3 (5) 29( 8) −22.7(8) −10.2(8) 2.3( 5) 48.0(1) 800 −23.5(4) −9.8(4) 3.4(4) 19 (2) 50(14) −20.6(4) −8.2(4) 3.8( 7) 61.5(2) 800 −23.9(5) −9.3(2) 3.4(2) 11.6 (4) 43( 4) −21.2(4) −7.8(4) 3.4( 4) 67.7(4) 800 −21.5(5) −8.1(2) 3.7(2) 8.9 (4) 42( 4) −21.7(5) −9.3(4) 3.4( 4) 71.6(2) 400 −22.8(2) −9.0(1) 4.0(2) 7.8 (2) 18( 3) −21.3(3) −6.2(2) 5 ( 3) 71.6(5) 800 −21.8(5) −8.4(2) 3.3(1) 7.3 (3) 42( 3) −21.1(5) −8.4(5) 3.7( 5) 74.6(2) 400 −22.9(1) −9.8(1) 3.6(1) 6.05(8) 19( 3) −29.0(2) −12.4(3) 2.8( 8) 74.5(4) 800 −22.6(3) −9.1(1) 3.7(1) 8.1 (1) 23( 2) −20.7(4) −12.4(3) 10 ( 8) 78.5(3) 550 −22.8(1) −9.0(1) 3.7(1) 7.8 (1) 18( 2) −19.8(3) − 7.2(8) 13 ( 6) 79.2(4) 400 −22.3(2) −8.8(1) 3.1(1) 6.6 (3) 3( 3) −26 (5) −13 (6) 15 ( 7) 79.8(8) 600 −23.6(4) −9.6(2) 3.4(2) 7.7 (3) 10( 3) −20 (5) −13 (6) 15 (13)
to deteriorate only after several hours. For example an experiment performed at = 26◦, νr = 800 Hz
showed the greatest order (µ = 7.3 ± 0.3◦) after 1–
2 h spinning. After 10 hours the order was reduced to µ = 10.6◦± 0.6◦. The contribution of the powder
phase is around 10% for both spectra and does not show a tendency to increase with time. For closer to the magic angle m the phases are less stable; it
takes a longer time to reach the most ordered state and the order is lost more quickly. For = 61.5◦the greatest order (µ = 11.6◦± 0.4◦) is reached after 10 h
spinning. Three hours later, the order had decreased to µ = 22◦± 1◦. The resonance position of the central
lines in both of the phases does not change with time, indicating thatδDand ¯β are not affected by prolonged spinning. All values reported in Table 1 and Figure 6 correspond to the state with the lowestµ.
At higher spinning frequencies (above 850 Hz), a change in the phase behavior is encountered. We have accelerated a sample, spun at = 76.5◦, from νr= 410 Hz to νr= 1840 Hz (Figure 8). At the lower
spinning frequency the sample behaved as described above, with an oriented phase at ¯β = 0◦ with µ = 6.6◦± 0.2◦,δDDMPC = −22.6◦± 0.2 ppm, δDDHPC =
Figure 8. (a)31P NMR experimental spectrum at = 76.5◦ with νr = 410 Hz; (b) best fit; (c) experimental spectrum at νr = 1840 Hz; (d) best fit. Note that the spinning-sideband pat-tern in spectrum (a) is not perfectly reproduced by the fit in (b), due to experimental instability in spinning rate which broadens the sidebands.
−9.1 ± 0.2 ppm and a randomly oriented phase with 23± 3% abundance (Figure 8a). After spinning 12 hours atνr= 1840 Hz, the spectrum of Figure 8c was
recorded. This spectrum contains two oriented phases: one phase with 63±25% abundance and ¯β = 90◦with δD
DMPC= −21.6±0.1 ppm, δ D
DHPC= −8.7±0.1 ppm,
q= 2.8 ± 0.2 and µ = 21◦± 4◦and another phase withβ = 0◦,δDDMPC = −22.7 ± 0.3 ppm, δDDHPC = −9.2 ± 0.3 ppm, q = 3± 1 and µ = 28◦ ± 7◦.
This indicates that, for higher spinning rates, a new oriented bicellar phase appears with the director per-pendicular to the spinning axis. This observation is in accordance with 2H NMR experiments at = 79◦ and νr = 920 Hz which also indicate that an
ori-ented phase with ¯β = 90◦appears for faster spinning (Zandomeneghi et al., 2001). It should however be noted, that the mosaic spread of both oriented phases is significantly higher than observed for slower spinning. It is possible that the same, not fully understood forces that stabilize the ¯β = 90◦phase for fast spin-ning are responsible for the transient order observed when the sample is spun at the magic angle.
Conclusions
Under variable-angle sample-spinning at frequencies below about 800 Hz, liquid-crystalline bicelle phases behave like ‘ordinary’ nematic liquid crystals with
χ < 0: the director of the dominant phase is
perpen-dicular to the rotation axis when the sample is spun at angles smaller than the magic angle and it orients parallel to the rotation axis when the angles are bigger. When the rotation angle is not near the magic angle, the phase is well ordered with a mosaic spreadµ of around 7◦. As the rotation angle approaches the magic angle, a second phase of increasing abundance is noted. This phase is much less ordered and can be well described by a powder distribution (no orientational order with respect to the magnetic field). At higher spinning frequencies two partially ordered phases have been detected at = 76.6◦. They have their direc-tors parallel and perpendicular to the spinning axis, respectively.
Our findings show that bicelles in rotating sam-ples are a promising medium for the investigation of membrane-bound peptides and proteins. The mosaic spread is comparable to the one in static samples or to the one of bilayers deposited on glass plates. The conditions can be optimized such that the disordered phase does not contribute significantly to the spec-trum. Notably, for angles close to = 90◦a highly ordered state can be created and maintained. From this initial state, a switched-angle spinning experiment
can prepare bicelles with the director parallel to the applied field. In addition, we propose an experiment where magic-angle-spinning spectra are obtained us-ing SAS techniques with the sample rotatus-ing at the magic angle, or close to it, only for the short time required to pulse and collect the NMR signal (see companion paper, Zandomeneghi et al., 2003).
Acknowledgements
We are grateful to Matthias Ernst, Megan Spence, Urban Meier and Andreas Hunkeler for helpful dis-cussion and for experimental assistance.
References
Bayle, J.P., Perez, F. and Courtieu, J. (1988) Liq. Cryst., 3, 753–758. Cheng, V.B., Suzukawa, H.H. and Wolfsberg, M. (1973) J. Chem.
Phys., 59, 3992–3999.
Courtieu, J., Alderman, D.W., Grant, D.M. and Bayles, J.P. (1982)
J. Chem. Phys., 77, 723–730.
Courtieu, J., Bayle, J.P. and Fung, B.M. (1994) Prog. Nucl. Magn.
Reson. Spectrosc., 26, 141–169.
Dufourc, E.J., Mayer, C., Stohrer, J., Althoff, G. and Kothe, G. (1992) Biophys. J., 61, 42–57.
Gaemers, S. and Bax, A. (2001) J. Am. Chem. Soc., 123, 12343– 12352.
Glaubitz, C. and Watts, A. (1998) J. Magn. Reson., 130, 305–316. Glover, K.J., Whiles, J.A., Wood, M.J., Melacini, G., Komives, E.A.
and Vold, R.R. (2001) Biochemistry, 40, 13137–13142. Hediger, S., Signer, P., Tomaselli, M., Ernst, R.R. and Meier, B.H.
(1997) J. Magn. Reson., 125, 291–301.
Howard, K.P. and Opella, S.J. (1996) J. Magn. Reson. Ser., B112, 91–94.
Kohler, S.J. and Klein, M.P. (1977) Biochemistry, 16, 519–526. Losonczi, J.A. and Prestegard, J.H. (1998) Biochemistry, 37, 706–
716.
Marassi, F.M., Opella, S.J., Juvvadi, P. and Merrifield, R.B. (1999)
Biophys. J., 77, 3152–3155.
Mehring, M. (1983) Principles of High Resolution NMR in Solids, Springer-Verlag, Berlin.
Moll, F. and Cross, T.A. (1990) Biophys. J., 57, 351–362. Nieh, M.P., Glinka, C.J., Krueger, S., Prosser, R.S. and Katsaras, J.
(2001) Langmuir, 17, 2629–2638.
Nieh, M.P., Glinka, C.J., Krueger, S., Prosser, R.S. and Katsaras, J. (2002) Biophys. J., 82, 2487–2498.
Picard, F., Paquet, M.J., Levesque, J., Belanger, A. and Auger, M. (1999) Biophys. J., 77, 888–902.
Prosser, R.S., Hwang, J.S. and Vold, R.R. (1998) Biophys. J., 74, 2405–2418.
Sanders, C.R. and Landis, G.C. (1995) Biochemistry, 34, 4030– 4040.
Sanders, C.R. and Schwonek, J.P. (1992) Biochemistry, 31, 8898– 8905.
Sanders, C.R., Hare, B.J., Howard, K.P. and Prestegard, J.H. (1994)
Prog. Nucl. Magn. Reson. Spectrosc., 26, 421–444.
Saupe, A. (1964) Z. Naturforsch., 19a, 161–171. Seelig, J. (1978) Biochim. Biophys. Acta, 515, 105–140.
Smith, S.A., Levante, T.O., Meier, B.H. and Ernst, R.R. (1994) J.
Magn. Reson. Ser., A106, 75–105.
Struppe, J., Komives, E.A., Taylor, S.S. and Vold, R.R. (1998)
Biochemistry, 37, 15523–15527.
Tian, F., Losonczi, J.A., Fischer, M.W.F. and Prestegard, J.H. (1999)
J. Biomol. NMR, 15, 145–150.
Vold, R.R. and Prosser, R.S. (1996) J. Magn. Reson. Ser., B113, 267–271.
Zandomeneghi, G., Tomaselli, M., Van Beek, J.D. and Meier, B.H. (2001) J. Am. Chem. Soc., 123, 910–913.
Zandomeneghi, G., Williams, P.T.F., Hunkeler, A. and Meier, B.H. (2003) J. Biomol. NMR, 25, 125–132.
![Table 1. δ D values and ratio q = [DMPC]/[DHPC] for the oriented as well as for the powder phase, mosaic spread of the bicellar director’s orientation, µ and amount of the powder phase determined in VAS experiments where the sample is spun at angles with](https://thumb-eu.123doks.com/thumbv2/123doknet/14838212.623590/9.892.175.724.151.521/table-values-oriented-bicellar-director-orientation-determined-experiments.webp)